Learn how to quickly set up the Dogra open-source voice agents platform on your local machine using Docker. This guide walks you through the necessary prerequisites and the single command needed to launch the application stack.
Chapters
00:00
Introduction to Dogra
Overview of the open source voice agents platform.
00:15
Local Setup Overview
00:26
Additional Resources
00:53
Prerequisites
01:17
Running Installation
Transcript
00:00
hi, my name is Sabiha and I'm a software engineer at Dogra.
00:03
Dogra is an open source voice agents platform which is very quick to get started with and the entire process at the platform is pretty straightforward.
00:15
this video in particular is to set up Dogra on your local system using Docker, so that you can run the application on your machine.
00:26
In case you want to set up Dogra at a remote machine, maybe with or without a custom domain then there are separate videos for that.
00:36
If you want to clone the code base and customize the implementation as per your use case then there is another video for the dev workflow.
00:48
And if you want to know more about the features of the platform then there is a separate video for that as well.
00:53
For this tutorial what you need in place is basically two things.
00:59
One is to have Docker installed on your system and the other is a fresh instance of Terminal.
01:08
So if you don't have Docker on your machine then you can go through this prerequisites page.
01:13
everything that you need will be there.
01:17
And the only command that we'll be needing in this tutorial is this one.
01:22
so this is our documentation page.
01:24
And the command is to basically run the application, the entire application stack.
01:33
So I'll copy this, I'll go to my terminal, it's a fresh instance and I'll paste the command and run it.
01:40
And what it does is it's going to bring the entire application stack up.
01:47
so that's going to take a couple of minutes and while that happens I'll just briefly cover what this command is about.
01:56
So it the first thing it does is it's it downloads the docker compose YAML file from our GitHub repository on your local machine and this file is basically responsible for defining the containers, all the containers that is that are required to run the application.
02:17
So it's a multi container application.
02:19
So and in order to run the containers we we need some images.
02:24
So the images we are pulling from is from the GitHub registry.
02:31
So you can go through the documentation if you want.
02:34
we also have our images uploaded at the Docker Hub registry.
02:39
So Additionally there is a telemetry which is optional, you can disable it if you want.
02:44
And this pool always ensures that whenever you are running the command you are always Pulling the latest images from the registry.
02:51
The latest dogra images from the registry.
02:55
so I would recommend that you use this command instead of the regular docker compose command that you typically use for any other system.
03:06
So the containers are all up.
03:10
we will copy this URL and we will launch the application now on browser.
03:15
I'll just launch an incognito tab and I launch the application here.
03:23
So we first need to create an account.
03:35
So in case you are using multiple different accounts on your local application and you are facing some odds related issues I would recommend that you log out and then log in again.
03:47
so what you see here is a voice agent builder and it has, it needs very very minimal data that in the sense that it has just two UM3 data entry points.
04:03
one is the kind of call that you want this agent to handle.
04:07
the other is the use case that you're building for and a description that describes what your agent does.
04:16
So for this demo I'll keep it very simple.
04:19
The use case is HR screening and what it does is it screens interview candidates.
04:30
And that creates a voice agent for us.
04:33
Of course if you want your voice agent to be highly specific, do a lot of things then you should definitely make that description of yours in the voice agent builder as exhaustive and to include everything that you want your voice agent to have.
04:52
So the next step once the voice agent gets built by the system is to test it out.
04:58
We'll do that with a web call.
05:00
I'll quickly grant the audio permission and we can start the web call here.
05:17
Hi sam, this is sabiha.
05:34
hey, so before that where are you calling from?
05:50
So you can see that there's a lot of information here which I didn't really key in at the voice agent builder.
05:59
So it assumes and makes fake data for the demo purpose.
06:05
and this entire screen summarizes everything that went that happened during the call.
06:12
we were able to see live transcript here and there are some additional nuances about the call in terms of the reasoning delay, the function calls and no transitions that are happening.
06:25
You also have access to your preview transcript and recording.
06:30
You can also download them.
06:33
There's also the context that was collected during the call.
06:37
So and there is a lot going on on the screen and I would advise you go through the documentation to understand everything.
06:43
You can definitely customize the agent but for all of that you can check out the documentation.
06:49
Everything is described in great detail.
06:53
So this is the this is how you get started on your local machine.
06:58
in case you have any questions about the setup you're facing, any challenges.
07:04
I'd love to hear them.
07:05
You can join our Slack community here.
07:07
The community is pretty active.
07:09
You can also access our cloud hosted platform which you can find@app.dogra.com the documentation again is hosted@docs.dogra.com and the our GitHub repository is with the name Dograj Qdobra.
07:29
You can check out the code base, it's completely open sourced and you can reach us out for any support that you need.